BRUSSELS — European regulators on Thursday welcomed AI leader Anthropic’s decision to defer the launch of its newest model over concerns that it could enable large-scale cybersecurity breaches.
“Given the model’s potential for large-scale cyber risk, we welcome this staged rollout,” European Commission spokesperson Thomas Regnier told POLITICO.
Anthropic announced on Tuesday evening that it shared its latest AI tech with a newly-formed group of 12 cybersecurity firms after finding that the new Claude Mythos model outperformed most humans at “finding and exploiting software vulnerabilities.”
These and 40 other unnamed organizations have access to the model to build their cyber defense capabilities and to scan and stress-test their systems.
Anthropic is a developer of what is called a “general-purpose AI model,” and under EU artificial intelligence rules, it is supposed to ensure that its model has an “adequate level of cybersecurity protection.”
The company also signed up to an EU code of practice under which it is mandated to address “risks from enabling large-scale sophisticated cyberattacks.”
The European Commission’s AI Office is “in dialogue with Anthropic” on how to implement that, Regnier said, but he declined to disclose details of these conversations as they are “commercially sensitive.”
Europa Monitor is a seasoned News Editor with over a decade of experience in journalism. Known for his editorial expertise and commitment to accuracy, Europa Monitor leads teams to deliver high-quality news content.
